In order to accurately predict the transformer far-field noise and evaluate the path contribution

a far-field noise prediction and path contribution solving method based on the transfer path analysis is proposed. The radiating surfaces of the two side-by-side transformers are equated to several virtual sound sources

and a "source-path-receiver" model is established. The virtual sound source signals are extracted by multi-channel least squares using the frequency response function matrix and the sound pressure vector measured at the near-field microphones. The sound pressure spectra of six points in the far field and the virtual source contributions are then calculated using the transmission path analysis method. Compared to the experimental results

the prediction errors are less than 2 dB at the initial moment

and the subsequent prediction errors over time are less than 3.5 dB. By evaluating the path contribution

the radiating surfaces with the highest noise contribution of the two transformers to a given zone are identified. The far-field noise prediction and path contribution evaluation method proposed in this paper can provide accurate far-field noise prediction for transformers

which can be further used for accurate real-time noise prediction in the complained neighborhood

as well as providing a data base for subsequent noise prevention treatments.